Self-Driving Car Engineer Nanodegree¶
Deep Learning¶
Project: Build a Traffic Sign Recognition Classifier¶
In this notebook, a template is provided for you to implement your functionality in stages which is required to successfully complete this project. If additional code is required that cannot be included in the notebook, be sure that the Python code is successfully imported and included in your submission, if necessary. Sections that begin with 'Implementation' in the header indicate where you should begin your implementation for your project. Note that some sections of implementation are optional, and will be marked with 'Optional' in the header.
In addition to implementing code, there will be questions that you must answer which relate to the project and your implementation. Each section where you will answer a question is preceded by a 'Question' header. Carefully read each question and provide thorough answers in the following text boxes that begin with 'Answer:'. Your project submission will be evaluated based on your answers to each of the questions and the implementation you provide.
Note: Code and Markdown cells can be executed using the Shift + Enter keyboard shortcut. In addition, Markdown cells can be edited by typically double-clicking the cell to enter edit mode.
Step 0: Load The Data¶
# Load pickled data
import pickle
# TODO: Fill this in based on where you saved the training and testing data
training_file = "./data/train.p"
testing_file = "./data/test.p"
with open(training_file, mode='rb') as f:
train = pickle.load(f)
with open(testing_file, mode='rb') as f:
test = pickle.load(f)
X_train, y_train = train['features'], train['labels']
X_test, y_test = test['features'], test['labels']
import numpy as np
print("keys",train.keys())
print("No of Train Samples",np.shape(X_train) )
print("No of Test Samples", np.shape(X_test))
print(np.shape(y_train)) # Test set Dimensions
print(len(np.unique(y_train))) #
Step 1: Dataset Summary & Exploration¶
The pickled data is a dictionary with 4 key/value pairs:
'features'is a 4D array containing raw pixel data of the traffic sign images, (num examples, width, height, channels).'labels'is a 2D array containing the label/class id of the traffic sign. The filesignnames.csvcontains id -> name mappings for each id.'sizes'is a list containing tuples, (width, height) representing the the original width and height the image.'coords'is a list containing tuples, (x1, y1, x2, y2) representing coordinates of a bounding box around the sign in the image. THESE COORDINATES ASSUME THE ORIGINAL IMAGE. THE PICKLED DATA CONTAINS RESIZED VERSIONS (32 by 32) OF THESE IMAGES
Complete the basic data summary below.
### Replace each question mark with the appropriate value.
# TODO: Number of training examples
n_train = np.shape(X_train)[0]
# TODO: Number of testing examples.
n_test = np.shape(X_test)[0]
# TODO: What's the shape of an traffic sign image?
image_shape = np.shape(X_train[0])
# TODO: How many unique classes/labels there are in the dataset.
n_classes = len(np.unique(y_train))
print("Number of training examples =", n_train)
print("Number of testing examples =", n_test)
print("Image data shape =", image_shape)
print("Number of classes =", n_classes)
Visualize the German Traffic Signs Dataset using the pickled file(s). This is open ended, suggestions include: plotting traffic sign images, plotting the count of each sign, etc.
The Matplotlib examples and gallery pages are a great resource for doing visualizations in Python.
NOTE: It's recommended you start with something simple first. If you wish to do more, come back to it after you've completed the rest of the sections.
### Data exploration visualization goes here.
### Feel free to use as many code cells as needed.
import matplotlib.pyplot as plt
import matplotlib.cbook as cbook
import random
# Visualizations will be shown in the notebook.
# %matplotlib inline
# ********************************************************************************
# SAMPLE TRAINING SET FOR VISUALIZATION
# ********************************************************************************
# Sample 400 images from train dataset for Visualization
# will Sample 400 images across the Training set Partitions
# to get Overview of Training Set Distribution.
rows = 20
columns = 20
def partition_set(image_set, rows=20, columns=20):
"""
Returns 4 sets with rows*columns randomly sampled from 4 Quantiles
"""
n_set = len(image_set)
# Parition Indices into 4 Parts
indx_Q1, indx_Q2, indx_Q3, indx_Q4 = np.array_split(np.array(range(n_set)), 4)
return(
image_set[[random.sample(list(indx_Q1), rows*columns)]], # Sample 1st Quantile
image_set[[random.sample(list(indx_Q2), rows*columns)]], # Sample 2nd Quantile
image_set[[random.sample(list(indx_Q3), rows*columns)]], # Sample 3rd Quantile
image_set[[random.sample(list(indx_Q4), rows*columns)]], # Sample 4th Quantile
)
# Function to Display Images
def show_images(image_set, columns=20):
# Arrange Images in Grid
image_set = np.concatenate( image_set ) # Concatenate All Images, Default: vertical Axis
image_set = np.concatenate(
np.split(
image_set, columns
), 1 # concatenation axis - Horizontal Axis
)
# Show Images
plt.figure(figsize=(10,10)) # Canvas Size
plt.imshow(image_set)
plt.show()
# Sample Training Set
train_Q1, train_Q2, train_Q3, train_Q4 = partition_set(X_train)
VISUALIZE TRAINING SET¶
# ====== First 400 Images from Training Set =========
show_images(X_train[:10],5)
# ===== Random Sample Images From First Quantile =======
show_images(train_Q1)
# ===== Random Sample Images From Second Quantile ======
show_images(train_Q2)
# ====== Random Sample Images From Third Quantile =========
show_images(train_Q3)
# ====== Random Sample Images From Fourth Quantile ========
show_images(train_Q4)
TRAIN and TEST LABEL DISTRIBUTION¶
# **********************************************************************
# Plot Train and Test Set Label Distribution
# **********************************************************************
plt.hist(y_train,43,color="green",label="Training Labels")
plt.hist(y_test,43,color="blue",label="Test Labels")
plt.legend()
plt.show()
Step 2: Design and Test a Model Architecture¶
Design and implement a deep learning model that learns to recognize traffic signs. Train and test your model on the German Traffic Sign Dataset.
There are various aspects to consider when thinking about this problem:
- Neural network architecture
- Play around preprocessing techniques (normalization, rgb to grayscale, etc)
- Number of examples per label (some have more than others).
- Generate fake data.
Here is an example of a published baseline model on this problem. It's not required to be familiar with the approach used in the paper but, it's good practice to try to read papers like these.
NOTE: The LeNet-5 implementation shown in the classroom at the end of the CNN lesson is a solid starting point. You'll have to change the number of classes and possibly the preprocessing, but aside from that it's plug and play!
Implementation¶
Use the code cell (or multiple code cells, if necessary) to implement the first step of your project. Once you have completed your implementation and are satisfied with the results, be sure to thoroughly answer the questions that follow.
### Preprocess the data here.
### Feel free to use as many code cells as needed.
from sklearn.utils import shuffle
X_train, y_train = shuffle(X_train, y_train)
VISUALIZE SHUFFLED TRAINING SET IMAGES¶
# Sample Training Set
train_Q1, train_Q2, train_Q3, train_Q4 = partition_set(X_train)
# ====== First 400 Images from Training Set =========
show_images(X_train[:400])
# ===== Random Sample Images From First Quantile =======
show_images(train_Q1)
# ===== Random Sample Images From Second Quantile ======
show_images(train_Q2)
# ====== Random Sample Images From Third Quantile =========
show_images(train_Q3)
# ====== Random Sample Images From Third Quantile =========
show_images(train_Q4)
Question 1¶
Describe how you preprocessed the data. Why did you choose that technique?
Answer: Visualizing samples from different sections of dataset, gives an impression,
1. certain category of signs concentrated to some sections.
2. Results a biased Validation set.
3. Training might take more iterations to converge.### Generate data additional data (OPTIONAL!)
### and split the data into training/validation/testing sets here.
### Feel free to use as many code cells as needed.
from sklearn.cross_validation import train_test_split
# Create Validation Data
print(np.shape(X_train))
print(np.shape(y_train))
X_train , X_valid, y_train, y_valid = train_test_split( X_train, y_train, test_size=0.15, random_state=42)
print(np.shape(X_train))
print(np.shape(y_train))
# Update No.Of Training samples
n_train = np.shape(X_train)[0]
Plot Train , Validation and Test Set Distribution¶
# Plot Train , Validation and Test Set Distribution
plt.hist(y_train,43,color="green",label="Training Samples")
plt.hist(y_test,43,color="blue",label="Test Samples")
plt.hist(y_valid,43,color="red",label="Validation Sample")
plt.legend()
VISUALIZE VALIDATION SET¶
# Sample Training Set
rows = 7
columns = 7
valid_Q1, valid_Q2, valid_Q3, valid_Q4 = partition_set(X_valid,rows,columns)
# ====== First 49 Images from Validation Set =========
show_images(X_valid[:49],columns)
# ===== VALIDATION SET : Random Sample Images From First Quantile =======
show_images(valid_Q1, columns)
# ===== VALIDATION SET : Random Sample Images From Second Quantile ======
show_images(valid_Q2,columns)
# ====== VALIDATION SET : Random Sample Images From Third Quantile =========
show_images(valid_Q3, columns)
# ====== VALIDATION SET : Random Sample Images From Third Quantile =========
show_images(valid_Q4,columns)
Question 2¶
Describe how you set up the training, validation and testing data for your model. Optional: If you generated additional data, how did you generate the data? Why did you generate the data? What are the differences in the new dataset (with generated data) from the original dataset?
Answer:
1. Training Data : 70% of Whole Data
2. Validation Data : 10% of Whole Data
3. Test Data : 20% of Whole Data### Define your architecture here.
### Feel free to use as many code cells as needed.
Input¶
The LeNet architecture accepts a 32x32xC image as input, where C is the number of color channels. Since MNIST images are grayscale, C is 1 in this case.
Architecture
Layer 1: Convolutional. The output shape should be 28x28x6.
Activation. Your choice of activation function.
Pooling. The output shape should be 14x14x6.
Layer 2: Convolutional. The output shape should be 10x10x16.
Activation. Your choice of activation function.
Pooling. The output shape should be 5x5x16.
Flatten. Flatten the output shape of the final pooling layer such that it's 1D instead of 3D. The easiest way to do is by using tf.contrib.layers.flatten, which is already imported for you.
Layer 3: Fully Connected. This should have 120 outputs.
Activation. Your choice of activation function.
Layer 4: Fully Connected. This should have 84 outputs.
Activation. Your choice of activation function.
Layer 5: Fully Connected (Logits). This should have 10 outputs.
Output
Return the result of the 2nd fully connected layer.
from tensorflow.contrib.layers import flatten
def LeNet(x,dropout):
"""
x : Batch of Images
dropout : Keep_prob
"""
global n_classes
# Hyperparameters
mu = 0
sigma = 0.1
# SOLUTION: Layer 1: Convolutional. Input = 32x32x1. Output = 28x28x18.
conv1_W = tf.Variable(tf.truncated_normal(shape=(5, 5, 3, 18), mean = mu, stddev = sigma))
conv1_b = tf.Variable(tf.zeros(18))
conv1 = tf.nn.conv2d(x, conv1_W, strides=[1, 1, 1, 1], padding='VALID') + conv1_b
# SOLUTION: Activation.
conv1 = tf.nn.relu(conv1)
# Dropout
conv1 = tf.nn.dropout(conv1, dropout)
# SOLUTION: Pooling. Input = 28x28x18. Output = 14x14x18.
conv1 = tf.nn.max_pool(conv1, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='VALID')
# SOLUTION: Layer 2: Convolutional. Output = 10x10x48.
conv2_W = tf.Variable(tf.truncated_normal(shape=(5, 5, 18, 48), mean = mu, stddev = sigma))
conv2_b = tf.Variable(tf.zeros(48))
conv2 = tf.nn.conv2d(conv1, conv2_W, strides=[1, 1, 1, 1], padding='VALID') + conv2_b
# SOLUTION: Activation.
conv2 = tf.nn.relu(conv2)
# Dropout
conv2 = tf.nn.dropout(conv2, dropout)
# SOLUTION: Pooling. Input = 10x10x48. Output = 5x5x48.
conv2 = tf.nn.max_pool(conv2, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='VALID')
# SOLUTION: Flatten. Input = 5x5x48. Output = 1200.
fc0 = flatten(conv2)
# SOLUTION: Layer 3: Fully Connected. Input = 1200. Output = 1024.
fc1_W = tf.Variable(tf.truncated_normal(shape=(1200, 1024), mean = mu, stddev = sigma))
fc1_b = tf.Variable(tf.zeros(1024))
fc1 = tf.matmul(fc0, fc1_W) + fc1_b
# SOLUTION: Activation.
fc1 = tf.nn.relu(fc1)
fc1 = tf.nn.dropout(fc1, dropout)
# SOLUTION: Layer 4: Fully Connected. Input = 1024. Output = 1024.
fc2_W = tf.Variable(tf.truncated_normal(shape=(1024, 1024), mean = mu, stddev = sigma))
fc2_b = tf.Variable(tf.zeros(1024))
fc2 = tf.matmul(fc1, fc2_W) + fc2_b
# SOLUTION: Activation.
fc2 = tf.nn.relu(fc2)
fc2 = tf.nn.dropout(fc2, dropout)
# SOLUTION: Layer 5: Fully Connected. Input = 1024. Output = 43.
fc3_W = tf.Variable(tf.truncated_normal(shape=(1024, n_classes), mean = mu, stddev = sigma))
fc3_b = tf.Variable(tf.zeros(n_classes))
logits = tf.matmul(fc2, fc3_W) + fc3_b
return logits
Features and Labels¶
Train LeNet to classify Traffic Sign data.
x is a placeholder for a batch of input images. y is a placeholder for a batch of output labels.
You do not need to modify this section.
import tensorflow as tf
x = tf.placeholder(tf.float32, (None, 32, 32, 3))
y = tf.placeholder(tf.int32, (None))
one_hot_y = tf.one_hot(y, n_classes)
dropout = tf.placeholder(tf.float32)
Training Pipeline¶
Create a training pipeline that uses the model to classify Traffic-Sign data.
You do not need to modify this section.
rate = 0.001
logits = LeNet(x,dropout)
cross_entropy = tf.nn.softmax_cross_entropy_with_logits(logits, one_hot_y)
loss_operation = tf.reduce_mean(cross_entropy)
optimizer = tf.train.AdamOptimizer(learning_rate = rate)
training_operation = optimizer.minimize(loss_operation)
Question 3¶
What does your final architecture look like? (Type of model, layers, sizes, connectivity, etc.) For reference on how to build a deep neural network using TensorFlow, see Deep Neural Network in TensorFlow from the classroom.
Answer:
### Train your model here.
### Feel free to use as many code cells as needed.
Model Evaluation¶
Evaluate how well the loss and accuracy of the model for a given dataset.
You do not need to modify this section.
correct_prediction = tf.equal(tf.argmax(logits, 1), tf.argmax(one_hot_y, 1))
accuracy_operation = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
saver = tf.train.Saver()
def evaluate(X_data, y_data):
"""
X_data : Images
y_data : labels
dropout : Keep_prob
"""
num_examples = len(X_data)
total_accuracy = 0
sess = tf.get_default_session()
for offset in range(0, num_examples, BATCH_SIZE):
batch_x, batch_y = X_data[offset:offset+BATCH_SIZE], y_data[offset:offset+BATCH_SIZE]
accuracy = sess.run(accuracy_operation, feed_dict={x: batch_x, y: batch_y, dropout:1.0}) # images, labels, keep_prob
total_accuracy += (accuracy * len(batch_x))
return total_accuracy / num_examples
Train the Model¶
Run the training data through the training pipeline to train the model.
Before each epoch, shuffle the training set.
After each epoch, measure the loss and accuracy of the validation set.
Save the model after training.
You do not need to modify this section.
import cv2
X_train_norm = []
X_valid_norm = []
X_test_norm = []
# Training set
for indx in range(len(X_train)):
image = X_train[indx]
image[:,:,0] = cv2.equalizeHist(image[:,:,0])
image[:,:,1] = cv2.equalizeHist(image[:,:,1])
image[:,:,2] = cv2.equalizeHist(image[:,:,2])
image = image/255. - 0.5
X_train_norm.append(image)
# Validation set
for indx in range(len(X_valid)):
image = X_valid[indx]
image[:,:,0] = cv2.equalizeHist(image[:,:,0])
image[:,:,1] = cv2.equalizeHist(image[:,:,1])
image[:,:,2] = cv2.equalizeHist(image[:,:,2])
image = image/255. - 0.5
X_valid_norm.append(image)
# Test set
for indx in range(len(X_test)):
image = X_test[indx]
image[:,:,0] = cv2.equalizeHist(image[:,:,0])
image[:,:,1] = cv2.equalizeHist(image[:,:,1])
image[:,:,2] = cv2.equalizeHist(image[:,:,2])
image = image/255. - 0.5
X_test_norm.append(image)
EPOCHS = 50
BATCH_SIZE = 128
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
num_examples = len(X_train)
print("Training...")
print()
for i in range(EPOCHS):
X_train_norm, y_train = shuffle(X_train_norm, y_train)
for offset in range(0, num_examples, BATCH_SIZE):
end = offset + BATCH_SIZE
batch_x, batch_y = X_train_norm[offset:end], y_train[offset:end]
sess.run(training_operation, feed_dict={x: batch_x, y: batch_y, dropout:1.0})
validation_accuracy = evaluate(X_valid_norm, y_valid) # Valid_images, Valid_Labels
print("EPOCH {} ...".format(i+1))
print("Validation Accuracy = {:.3f}".format(validation_accuracy))
print()
saver.save(sess, 'lenet')
print("Model saved")
# MODEL TESTING¶
with tf.Session() as sess:
saver.restore(sess, tf.train.latest_checkpoint('.'))
test_accuracy = evaluate(X_test_norm, y_test) # test_images, test_labels, keep_prob
print("Test Accuracy = {:.3f}".format(test_accuracy))
METRICS¶
# ====== Generate Logits for Test Data ======
with tf.Session() as sess:
saver.restore(sess, tf.train.latest_checkpoint('.'))
lgts = sess.run(logits, feed_dict={x:X_test, dropout:1.0})
VISUALIZE SAMPLE IMAGES FOR EACH CLASS¶
# Sample 4 images from each class for visualization.
for img_class in range(n_classes):
images_class = X_train[np.equal(y_train,img_class)] # Gather images of a Class
images_class = images_class[np.random.choice(range(len(images_class)), 4)] # pick 4 random images
if img_class == 0:
image_classes = images_class[::]
continue
image_classes = np.concatenate((image_classes, images_class))
image_classes = np.concatenate((image_classes, images_class)) # Duplicate last Class to make even no of classes
show_images(image_classes,22) # image, columns
CONFUSION MATRIX¶
import itertools
np.set_printoptions(threshold=np.inf)
# Calculate Confusion Matrix
cm = np.zeros([n_classes,n_classes])
for i, j in zip( np.argmax(lgts,1), y_test ):
cm[i][j] += 1
# Compute % of miss/correct classification
cm_csum = np.sum(cm,axis=1)
cm_per = np.divide(cm, cm_csum)*100
# Save Confusion matrix
np.savetxt("ConfusionMatrix",cm,delimiter=",")
# Generate Confusion Matrix map
plt.figure(figsize=(20,10))
plt.pcolor(cm_per)
plt.ylabel("Predicted Label")
plt.xlabel("Given Label")
plt.colorbar()
plt.show()
# A PEEK @ Test Image Vs Classified Class (for missclassified images)¶
y_pred = np.argmax(lgts,1)
false_pred_test_images = X_test[np.not_equal(y_pred, y_test)]
false_pred_labels = y_pred[np.not_equal(y_pred, y_test)]
sample_fal_pred_test_images = false_pred_test_images[:20]
sample_fal_pred_labels = false_pred_labels[:20]*4
print("Test-Image","\t","Classified Class")
plt.figure(figsize=(10,30))
plt.imshow(
np.concatenate(
[np.concatenate(sample_fal_pred_test_images),
np.concatenate(image_classes[sample_fal_pred_labels])],
axis=1
)
)
plt.show()
CLASSIFICATION SUMMARY By CLASS¶
# Plot MissClassified Symbol Histogram
plt.hist(y_train, 43, color="red",label="Training Samples")
plt.hist(y_test, 43, color="green",label="Testing Samples")
plt.hist(y_test[np.not_equal(np.argmax(lgts,1), y_test)] , 43, label="Missclassified test Samples")
plt.legend()
plt.show()
# Plot Correctly Classified Symbol Histogram
plt.hist(y_train, 43, color="red",label="Training Samples")
plt.hist(y_test, 43, color="green",label="Testing Samples")
plt.hist(y_test[np.equal(np.argmax(lgts,1), y_test)] , 43, label="Correctly classified test Samples")
plt.legend()
plt.show()
Question 4¶
How did you train your model? (Type of optimizer, batch size, epochs, hyperparameters, etc.)
Answer: Model :
1. Started with Basic LeNet Architecture
2. 3-fold increase in filters to accomodate more features
3. 50% DropOut to compensate for overfitting
4. Batch size : 128
5. Ran for 50 epochs
6. with 0.001 Learning rateQuestion 5¶
What approach did you take in coming up with a solution to this problem? It may have been a process of trial and error, in which case, outline the steps you took to get to the final solution and why you chose those steps. Perhaps your solution involved an already well known implementation or architecture. In this case, discuss why you think this is suitable for the current problem.
Answer:
Started with Basic LeNet Architecture. confusion matrix for test set shows localized misclassification patterns like,
most of missclassified speed limits signs are confused against other speed limits rather caution triangle or privilage
Diamond signs. Another interesting pattern observed is speed limits are missclassified against direction signs, which
shows circles/triangle features are strongly classified. observations shows similarity of features like pedestrian/caution
caution/lane merging require more features at first stage conv layer for better classification.
I choose to increase features with 50% dropout to compensate for overfitting. Also Normalize brightness and scale pixel values
0.5 to -0.5. with these I see 95.5% test set accuracy against, 90% test set accuracy with basic LeNet Architecture.
with current test set accuracy is 3% off from human error. I feel data augmentation might help for better accuracy numbers, whic
I couldnt implement in the interest of time, would to consider data augmentation for future work.
since LeNet uses 5x5 feature map, also like to check the performance of 3x3 feature maps.Step 3: Test a Model on New Images¶
Take several pictures of traffic signs that you find on the web or around you (at least five), and run them through your classifier on your computer to produce example results. The classifier might not recognize some local signs but it could prove interesting nonetheless.
You may find signnames.csv useful as it contains mappings from the class id (integer) to the actual sign name.
Implementation¶
Use the code cell (or multiple code cells, if necessary) to implement the first step of your project. Once you have completed your implementation and are satisfied with the results, be sure to thoroughly answer the questions that follow.
### Load the images and plot them here.
### Feel free to use as many code cells as needed.
from PIL import Image
import glob
def PIL2array(img):
return np.array(img.getdata(),
np.uint8).reshape(img.size[1], img.size[0], 3)
web_test = [] # Init Empty Collection
for img_png in glob.glob('data/web/dataset/*.png'): #assuming png
print(img_png)
png=Image.open(img_png)
png.thumbnail((32,32), Image.ANTIALIAS) # Resize to 32x32
# Convert to RGB
png.load()
bg = Image.new("RGB", png.size, (255, 255, 255))
bg.paste(png, mask=png.split()[3]) # 3 is the alpha channel
print(np.shape(bg))
web_test.append(PIL2array(bg))
#web_test.append(bg)
plt.imshow(np.concatenate(web_test))
print(np.shape(web_test))
plt.show()
Question 6¶
Choose five candidate images of traffic signs and provide them in the report. Are there any particular qualities of the image(s) that might make classification difficult? It could be helpful to plot the images in the notebook.
Answer: I collected German Traffic signs from wiki page, which has better quality images.
Challenges :
1. size of image : Images collected from internet need to be scaled to 32x32 size for better test results.### Run the predictions here.
### Feel free to use as many code cells as needed.
# MODEL TESTING¶
# Pre-Process web Data
web_test_norm = []
# Training set
for indx in range(len(web_test)):
image = web_test[indx]
image[:,:,0] = cv2.equalizeHist(image[:,:,0])
image[:,:,1] = cv2.equalizeHist(image[:,:,1])
image[:,:,2] = cv2.equalizeHist(image[:,:,2])
image = image/255. - 0.5
web_test_norm.append(image)
with tf.Session() as sess:
saver.restore(sess, tf.train.latest_checkpoint('.'))
lgts = sess.run(logits, feed_dict={x:web_test_norm, dropout:1.0})
print(np.argmax(lgts,1))
Question 7¶
Is your model able to perform equally well on captured pictures when compared to testing on the dataset? The simplest way to do this check the accuracy of the predictions. For example, if the model predicted 1 out of 5 signs correctly, it's 20% accurate.
NOTE: You could check the accuracy manually by using signnames.csv (same directory). This file has a mapping from the class id (0-42) to the corresponding sign name. So, you could take the class id the model outputs, lookup the name in signnames.csv and see if it matches the sign from the image.
Answer: I see model performace is equal/better to what I observed with test set. Reasons for better test results, I would think of lucky enough to get good quality pics from internet
lgts_k = tf.nn.top_k(lgts[:6,:], k=5)
with tf.Session() as ses:
lgts_k5 = ses.run(lgts_k)
# Preview top 5 probabilities
print(np.shape(lgts_k5))
print(lgts_k5)
# Visualize logits as color map
plt.pcolor(lgts[1:20])
plt.colorbar()
plt.show()
Question 8¶
Use the model's softmax probabilities to visualize the certainty of its predictions, tf.nn.top_k could prove helpful here. Which predictions is the model certain of? Uncertain? If the model was incorrect in its initial prediction, does the correct prediction appear in the top k? (k should be 5 at most)
tf.nn.top_k will return the values and indices (class ids) of the top k predictions. So if k=3, for each sign, it'll return the 3 largest probabilities (out of a possible 43) and the correspoding class ids.
Take this numpy array as an example:
# (5, 6) array
a = np.array([[ 0.24879643, 0.07032244, 0.12641572, 0.34763842, 0.07893497,
0.12789202],
[ 0.28086119, 0.27569815, 0.08594638, 0.0178669 , 0.18063401,
0.15899337],
[ 0.26076848, 0.23664738, 0.08020603, 0.07001922, 0.1134371 ,
0.23892179],
[ 0.11943333, 0.29198961, 0.02605103, 0.26234032, 0.1351348 ,
0.16505091],
[ 0.09561176, 0.34396535, 0.0643941 , 0.16240774, 0.24206137,
0.09155967]])Running it through sess.run(tf.nn.top_k(tf.constant(a), k=3)) produces:
TopKV2(values=array([[ 0.34763842, 0.24879643, 0.12789202],
[ 0.28086119, 0.27569815, 0.18063401],
[ 0.26076848, 0.23892179, 0.23664738],
[ 0.29198961, 0.26234032, 0.16505091],
[ 0.34396535, 0.24206137, 0.16240774]]), indices=array([[3, 0, 5],
[0, 1, 4],
[0, 5, 1],
[1, 3, 5],
[1, 4, 3]], dtype=int32))Looking just at the first row we get [ 0.34763842, 0.24879643, 0.12789202], you can confirm these are the 3 largest probabilities in a. You'll also notice [3, 0, 5] are the corresponding indices.
Answer:
1. For Correctly Classified Images, I see peakiness in the probabilities.
2. For miss classified images, probabilities were smoother.Note: Once you have completed all of the code implementations and successfully answered each question above, you may finalize your work by exporting the iPython Notebook as an HTML document. You can do this by using the menu above and navigating to \n", "File -> Download as -> HTML (.html). Include the finished document along with this notebook as your submission.